Prediction of soil urea content using RF spectroscopy and partial least square regression

Abstract

Soil analysis plays a crucial role in increasing the crop yield and in the conservation of soil resources. Conventional methods of soil analysis are expensive, time consuming and do not take into account the spatial and temporal variability of a field which can lead to over or under application of fertilizers. Thus, in the long run the soil loses its fertility and also causes ground water contamination. Hence, there is a need for in-situ measurement of soil parameters that takes into account the spatial and temporal variability of the field. The objective of this study is to predict soil urea content using RF spectroscopy and partial least square regression. Around 20 samples having different concentrations of urea, sodium chloride, calcium carbonate, potash and phosphate were taken for the study. Out of which 16 samples were used for calibration and 4 samples were used for prediction purpose. The coefficient of determination R2 obtained was 0.987 and RMSE was found to be 3.183 for urea which corresponds to an error rate of 1.84 percent.